Proceedings of the International Conference on Industrial Engineering and Operations Management

Pilsen, Czech Republic, July 23-26, 2019

An Approach of Designing Robust Plant Layout Using

Genetic Algorithm
Rahul Sakharwade, Udit Narayan Sahu, Mridul Singh Rajput
Department of Mechanical Engineering,
National Institute of Technology,
Raipur, Chhattisgarh, India
rahul.sakharwade279@gmail.com, Sahuudit09@gmail.com, msrajput.me@nitrr.ac.in

Abstract
In today’s competitive global environment, industries manufacture different types of products to fulfill market
demands. This results into a number of diversified processes to be performed in industries. These processes are
performed on different machines to produce a final product. So, a need arises for proper arrangement of machines in
the workplace. For this, plant layout is an effective tool for allocating different machines to a different location. The
main objective of plant layout is to provide smooth workflow throughout the facility. In this paper, a new model of
the process type of plant layout is provided with an aim of minimizing the total material flow cost by reducing the
unnecessary movement between the machines. The genetic algorithm is suggested to compute allocations of
machines and estimation of cost incurred in the layout. Finally, a numerical illustration is discussed for reducing the
material flow cost to prove the efficacy of the proposed model.

Keywords
Process layout; Plant layout problem; Genetic algorithm; Material handling cost

1. Introduction
The prime objective of any manufacturing industry is to maximize profit margins and at the same time, reduce costs.
So, the function of production plays a key role in the success of the organization. For achieving efficient production
and service systems, along with optimal planning and policies, there is a need of a well designed plant layout.
Optimal Design of plant layout plays an important role not only in the initial stages, but also has a great influence at
the later stage. It has a wide impact on productivity; work in process, manufacturing lead time, material handling
cost and on other areas of operation Pillai et al. (2011). In general, the plant layout problem is concerned with
reduction of material flow/handling cost. Since, material handling cost is one of the major costs incurred in
production, which can contribute up to 20-50% of operating costs Chan et al. (2004). Material handling is a non-
value adding activity. The efficiency of a plant can be increased by reducing such non-value activities.

Plant layout is a study which deals with the location and arrangement of different machines/facilities within a
stipulated area in order to achieve greatest coordination and efficiency of 4M’s (Men, Material, Machines and
Methods) in a plant. Layout planning is a complex activity involving optimization of positions of machines and
workstations. The objective of a good plant layout is to minimize material handling and movement of work from one
department to the next.

There are different types of plant layout which are applicable in different industries according to their requirement.
In this study, the problem is considered based on the process type of plant layout. The process layout can
accommodate a variety of products. This requires a greater flexibility in production, thus allowing industries to
fulfill a wide range of customer’s demands. In this type of layout, machines are arranged according to the nature or
type of operations performed for manufacturing purpose. Each product has a definite sequence of operations, not

© IEOM Society International 763

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

necessarily in the same order in which the machines are placed. So, there is frequently non-unidirectional movement
of workflow takes place between the facilities, according to the changing demand of products. The objective of this
study is to reduce the material handling cost by assigning the machine/facility in an optimal way in each location
and hence, increase productivity.

Static plant layout (SPL) problem approach is suited for facility layout problems when the demand is more or less
constant with time. But, when the demand is varied with respect to time, then, dynamic plant layout approach is
more efficient for various planning horizons. There are two major approaches to solve dynamic plant layout
problem. One approach is adaptive approach which assumes that the machines can be easily relocated from time to
time with low rearrangement costs. On the other hand, the robust approach focuses on designing a layout with
minimum the total material handling costs by assuming that the rearrangement cost as too high Pillai et al. (2011).
Generally, in practice the rearrangement cost is very high in industries due to bulk movement of machines and
complex rearrangement process. So, this study suggests a robust model for dynamic plant layout problem, by using
genetic algorithm for layout formation. The remainder of this study is organized as follows. Section 2 discussed the
concerned literature while problem formulation is provided in section 3. Numerical illustration along with the results
is discussed in section 4. Finally, section 5 provides the concluding remark of this study.

2. Literature Review
With growing competition of products in the market, industrialist and researchers are trying to fulfill the customer
demands by making it available at lower prices. One contributor to these prices is material handling costs. This cost
can be reduced through proper design of plant layout. Previously, many researchers have contributed their efforts to
develop methods for designing plant layout. Pillai et al. (2011) designed a robust facility layout by considering
dynamic market demand. In their work, applied simulated annealing (SA) approach to a case study from the work of
Chan et al. (2004) and obtained improved results for the same case study. They also used total penalty cost to test
the suitability of robust layout for the case considered. There are various approaches to design the plant layout as
well as to check its suitability to be a robust layout. Prasad et al. (2014) designed a manufacturing plant layout with
a goal of minimizing the material flow cost by positioning the machines within the existing area. In their work,
Standard for the Exchange of Product (STEP) file is considered as an input to Java program and output is obtained
as a distance matrix between the locations. The distance matrix is used to calculate the layout cost. Computerized
Relative Allocation of Facilities Technique (CRAFT) is then applied to design the plant layout. This results in
significant reduction in layout cost.

Bock and Hoberg (2007) introduced a detailed layout planning by considering simultaneously both machine
arrangements and transportation path. In their work, the analysis was performed by incorporating specific attributes
to certain layout subareas and based on computational experiments; they developed a mathematical model which
showed that the elaborate heuristics generate better layout configurations. Neghabi et al. (2014) focused on the
effectiveness of a robust approach to solve facility layout problem (FLP). The authors have considered uncertainty
in the size of the department length and for modeling this uncertainty; the department size is considered as a
bounded variable and can be changed by using an agent based simulation model (ABSMODEL).

Shahin and Poormostafa (2011) proposed facility layout design improvement and optimization by using an
integrated approach of simulation, quality function deployment and fuzzy analytic hierarchy process. Mishra et al.
(2011) described the shifting cost of the machine and also use replacement analysis technique for redesigning the
plant layout. The authors employed sampling technique for solving large size problem. Earlier researchers
Balakrishnan et al. (2009), Balakrishnan et al. (2000), Corry et al. (2004), and Kulturel-Konak, S. (2007) had given
flexible or adaptive or agile layout which can rearrange the facilities for the purpose of meeting the requirements of
production.

© IEOM Society International 764

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

A systematic layout planning theory for reducing the production cost is described by Hossain et al. (2015) where a
detailed study of the operation process chart and activity relationship chart were investigated. Lee (2015) formulated
a mathematical program for a multi floor plant layout problem. In this model, the author considered the safety
distance as well as impacts of physical explosion on pipelines under different condition. The main objective is to
minimize the pipeline cost for connecting equipments. The objective is achieved by employing a mixed integer
nonlinear programming (MINLP) solvers. The results showed a more reasonable and safer plant layout.
Palomo-Romero et al. (2017) proposed island model genetic algorithm to solve unequal area facility layout problem
and also maintained population diversity for obtaining wider sampling of the search space to get better quality
solutions in fewer generations.

Use of Genetic algorithm (GA) as a tool for optimizing the parameters is also described by various researchers.
Datta (2012) applied a proposed genetic algorithm to linear programming problems in finding the fittest
chromosomes. Genetic algorithm is gaining more importance for solving different types of problems. Belea (2004)
described and formulated the dynamic nature of genetic algorithm using the state equation by generalizing the
schema theorem. The use of the generic genetic algorithm was made by Cuibus and Letia (2012), to test the system
behavior and for the evaluation of control performances in a problem of lake system control.

Nicoara (2007) employed genetic algorithm for solving job shop scheduling problem in the pharmaceutical industry
and described the problem as both single-objective and multi-objective problem by considering the various
complexity. The results provided an efficient solution for the case study considered by the author. Halal and
Dumitrache (2006) described the power of genetic algorithm for navigation purpose by considering three different
cases of mobile robot’s behavior for designing the behavioral controllers. The results showed that GA has good
navigation ability.

In this paper, the genetic algorithm is used for designing process type plant layout problem with the aim of reducing
material handling cost.

3. Problem Description
The study is based on the process layout which is one of type of plant layout where the arrangements of machines
are made according to the type and nature of operations performed. In this type of layout, every work does not flow
through same lines, but follows a different sequence of operations for achieving the final product. The layout design
problem consists of assignment of machines in discrete locations with an aim of minimizing total material handling
cost. The assignment cost is considered as the material handling cost which is the product of the distances between
the locations and flow of materials between the department / machine. Different planning periods are considered for
involving the dynamic market environment. Planning period describes quantity of different products to be
manufactured according to the changing demand of customers. Based on these, material flow between the
departments or machines is calculated. The mathematical model is formulated to minimize the material handling
cost.

3.1 Assumptions of the model

Each facility location is considered as the coordinates of the department’s centre in initial layout. All distances
between the locations are calculated through Euclidean formula. The material flow cost for a unit product is
assumed as a unit cost per unit distance travel. All departments are of equal area.

3.2 Mathematical model for plant layout problem

n n
Minimize Z = ∑∑ DijFijSij
i =1 j =1 Eq. (I)

© IEOM Society International 765

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

Where, Z = total material handling costs.

Dij = distance between ith departments to jth departments.
Fij = material flow between ith departments to jth departments.
Sij = 1, i.e. Unit cost of travel of unit product for unit distance travel.
Where, Sii & Sjj = 0
For calculating the distance between the locations, a Euclidian distance formula is used Palomo-Romero (2017).
Distance = �(x2 − x1 )2 + (y2 − y1 )2 Eq. (II)
Where, (x1, y1) and (x2, y2) is the coordinates of locations.
The total minimum material handling cost incurred on allocating machine to each location is calculated by
n
Cij = ∑ Fij(max)Dij(min) Eq. (III)
i, j

Cij = Cost incurred for allocating machine to each location.

Fij(max) = Maximum flow between the machines.
Dij(min) = Minimum distance between the locations.

3.3 Proposed Genetic algorithm (GA) for layout formation

The genetic algorithm is an evolutionary optimization technique based on the ideas of natural selection and natural
genetics. It is based on the “survival of the fittest” concept and simulates on the process of evolution. For a wide
class of problem, GA gives better results Prasad et al (2014). GA works with initial solution, in this work, initial
layout is considered as the initial solution for the algorithm. By using initial solution, the new solution is generated
with an aim of improving the quality of the previous one. The operators such as selection, crossover and mutation
were used to create the new solution. On the basis of fitness function value, the survivors were decided for the next
generation.

The GA solver in the optimization toolbox describes stopping criteria which determine the causes to terminate the
algorithm. There are some stopping criteria listed below:
Max iterations: represents the maximum number of iterations the algorithm performs.
Max function evaluations: specifies the maximum number of evaluations the algorithm performs for objective and
constraint functions.
X tolerance: the termination tolerance for variable ‘x’.
Function tolerance: represents the termination tolerance for the objective function value.
Nonlinear constraint tolerance: specifies the tolerance for the maximum nonlinear constraint violation.
Unbounded threshold: terminates iterations if the objective function value lies below it.

3.4 Pseudo code of Genetic algorithm

Begin
{
Generate an initial population;
Evaluate the population fitness;
While (termination criteria not met)
{
Selection of parents for reproduction;
Perform crossover and mutation operation;
Evaluate the population fitness;
}
}

© IEOM Society International 766

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

3.5 Flow chart of Genetic algorithm

Figure 1. Flow chart of Genetic algorithm

4. Numerical Demonstration and Result Discussion:

The layout design problem consists of assignment of machines in discrete locations with the aim of minimizing the
total material handling cost. The assignment cost is considered as the material handling cost which is the product of
the distances between the locations and flow of materials between the department/machines.
The required data for the problem consists of a sequence of operations performed for different products, number of
products manufactured in different planning horizon and location layout grid. An arbitrary layout is considered in
evaluating the performance of layout formation method. The layout consists of six machines, i.e. MA, MB, MC, MD,
ME and MF located at six locations L1, L2, L3, L4, L5 and L6 respectively. The initial layout is considered as under.

Table 1. Initial plant Layout

L1 L2 L3 L4 L5 L6
MA 1 0 0 0 0 0
MB 0 1 0 0 0 0
MC 0 0 1 0 0 0
MD 0 0 0 1 0 0
ME 0 0 0 0 1 0
MF 0 0 0 0 0 1

These locations are generated arbitrarily in an area of (24X8) workspace in MATLAB software. The distance
between each location with respect to the other is calculated by using Euclidian distance formula. These distances
are represented through distance matrix shown in table 2. The sequence of machines employed in performing the
operations on four different products, i.e. P1, P2, P3, and P4 is considered and on that basis, workflow between the
machines is generated. For providing the dynamic market environment, dynamic demand of products is considered
by employing multiple planning periods. Five different planning periods are considered to formulate the layout

© IEOM Society International 767

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

design problem and is shown in table 4. Each planning period describes the quantity of four different types of
products manufactured.
Table 2. Distance matrix generated between locations

Locations L1 L2 L3 L4 L5 L6
L1 00.00 14.04 03.61 11.18 15.00 13.00
L2 14.04 00.00 17.03 04.25 01.41 01.41
L3 03.61 17.03 00.00 14.51 17.11 16.13
L4 11.18 04.25 14.56 00.00 04.47 02.82
L5 15.00 01.41 17.11 04.47 00.00 02.00
L6 13.00 01.41 16.13 02.82 02.00 00.00

Table 3. Operation sequence of products on machines

Products Operational sequence of machines

P1 MA - MB - M D - MB - MC - MF
P2 MA - MD - MC - MB - MD - MF
P3 MA - MB - ME - MC - MF
P4 MA - MC - M D - ME - MF

Table 4. Product demand (number of units) for the different planning period

Products demand for Planning periods

Products
1st period 2nd period 3rd period 4th period 5th period
P1 30 25 15 05 10
P2 25 40 20 20 25
P3 15 20 35 15 30
P4 20 10 25 45 15

From table 3 and table 4, work flow matrix between different departments is formed. The workflow depends upon
the sequence of operations and the number of products manufactured in each planning period.

Table 5. Work flow matrix considering all planning periods

FROM \ TO MA MB MC MD ME MF
MA 200 115 130
MB 85 215 115
MC 130 115 200
MD 85 130 115 130
ME 115 115
MF

Assume the cost of material handling for moving from one department to another as one unit per distance travel. The
cost directly depends upon the total distance travelled between the machines. The total minimum cost matrix is
generated on the basis of work flow matrix and distance matrix by using equation (III) for each allocation of
machine at each location.

© IEOM Society International 768

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

Table 6. Total minimum cost matrix

L1 L2 L3 L4 L5 L6
MA 3670.40 954.20 4469.75 1630.55 1056.10 866.30

MB 3255.25 826.85 3822.00 1475.00 913.00 797.20

MC 3670.40 954.20 4469.75 1630.55 1056.10 866.30

MD 4611.00 2048.12 5664.60 2383.45 2232.00 1872.00

ME 1700.00 324.30 2089.00 813.00 392.00 392.00

MF 0.00 0.00 0.00 0.00 0.00 0.00

The Genetic algorithm (GA) solver in MATLAB is employed to solve the model. Population size is considered as
200. There are 36 variables considered for the purpose of allocation and 100 values of each variable are generated
during iteration. Assuming a crossover fraction of 0.8 and considering all stopping criteria for termination of the
algorithm, the following result is obtained.

(I)

(II)

© IEOM Society International 769

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

(III)
Figure 2. (I) Best fitness chart; (II) Current best individual; (III) Selection function chart.

Table 7. Results obtained through GA solver

L1 L2 L3 L4 L5 L6
MA 0.52 0.05 0.13 0.05 0.20 0.05
MB 0.02 0.92 0.00 0.01 0.05 0.01
MC 0.02 0.01 0.00 0.91 0.05 0.01
MD 0.01 0.01 0.45 0.01 0.51 0.01
ME 0.01 0.01 0.00 0.01 0.05 0.92
MF 0.42 0.00 0.45 0.00 0.13 0.00

Based on the results shown in table 7, allocations of machines are made on the different locations. This allocation is
done where the variable has the higher proportionate value compared to the other variables in the respective row.

Table 8. Allocations made on the basis of GA results

L1 L2 L3 L4 L5 L6
MA 1 0 0 0 0 0
MB 0 1 0 0 0 0
MC 0 0 0 1 0 0
MD 0 0 0 0 1 0
ME 0 0 0 0 0 1
MF 0 0 1 0 0 0

Table 9. Comparison of initial layout cost and proposed layout cost

Planning period Initial layout cost Proposed layout cost % reduction in cost
1st period 2319.45 1769.75 23.67%
2nd period 2412.45 1739.55 27.89%
3rd period 2261.80 1708.05 24.48%
4th period 1951.25 1386.27 28.95%
5th period 1917.40 1399.70 27.00%
All planning periods 11742.15 8751.50 25.46%

© IEOM Society International 770

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

The total material handling cost is calculated based on the allocations for each of the planning period. The
corresponding percentage reduction in the proposed layout cost compared to the initial layout cost is obtained as
described in table 9.

The result shows that for a fourth planning period, there is the maximum reduction in cost, i.e. 28.95%, whereas, the
overall reduction in cost by considering all planning periods is achieved as 25.46%.

5. Conclusion:
In this work, a method for process type plant layout is proposed with the aim of reducing total material handling
cost. Since material handling cost is one of the major contributors in increasing the cost of production. A robust
layout procedure is developed for dynamic environment, where different demand scenarios are considered. Robust
layout is fixed for different planning periods. This is being clearly shown in numerical illustration. Genetic
algorithm (GA) solver is employed in MATLAB for optimizing the proposed process type plant layout. Further,
result shows that the proposed method is able to reduce material handling cost significantly. This proposed method
can further be applied to similar situations in industries with some or little modification. This work can be further
extended by considering numerous decisive factors for the cost and constraints of floor area for similar types of
small or large scale industries.

References

Pillai, V. M., Hunagund, I. B., & Krishnan, K. K. (2011). Design of robust layout for dynamic plant layout
problems. Computers & Industrial Engineering, 61(3), 813-823.
Chan, W. M., Chan, C. Y., & Kwong, C. K. (2004). Development of the MAIN algorithm for a cellular
manufacturing machine layout. International journal of production research, 42(1), 51-65.
Prasad, N. H., Rajyalakshmi, G., & Reddy, A. S. (2014). A Typical Manufacturing Plant Layout Design Using
CRAFT Algorithm. Procedia Engineering, 97, 1808-1814.
Bock, S., & Hoberg, K. (2007). Detailed layout planning for irregularly-shaped machines with transportation path
design. European Journal of Operational Research, 177(2), 693-718.
Neghabi, H., Eshghi, K., & Salmani, M. H. (2014). A new model for robust facility layout problem. Information
Sciences, 278, 498-509.
Shahin, A., & Poormostafa, M. (2011). Facility layout simulation and optimization: An integration of advanced
quality and decision making tools and techniques. Modern Applied Science, 5(4), 95.
Mishra, P. M., Khare, V. K., & Mishra, P. (2011). Cost Optimization of Plant Layout Problems by Matrix
Algorithm. International Journal, 4.
Balakrishnan, J., & Cheng, C. H. (2009). The dynamic plant layout problem: Incorporating rolling horizons and
forecast uncertainty. Omega, 37(1), 165-177.
Balakrishnan Chun Hung Cheng Dani, J. (2000). An improved pair-wise exchange heuristic for the dynamic plant
layout problem. International Journal of Production Research, 38(13), 3067-3077.
Corry, P., & Kozan, E. (2004). Ant colony optimization for machine layout problems. Computational optimization
and applications, 28(3), 287-310.
Kulturel-Konak, S. (2007). Approaches to uncertainties in facility layout problems: Perspectives at the beginning of
the 21st Century. Journal of Intelligent Manufacturing, 18(2), 273-284.
Hossain, R., Rasel, M. K., & Talapatra, S. (2015). Increasing Productivity through Facility Layout Improvement
using Systematic Layout Planning Pattern Theory. Global Journal of Research In Engineering, 14(7).
Lee, C. J. (2015). Optimal multi-floor plant layout based on the mathematical programming and particle swarm
optimization. Industrial health, 53(6), 491-497.
Palomo-Romero, J. M., Salas-Morera, L., & García-Hernández, L. (2017). An island model genetic algorithm for
unequal area facility layout problems. Expert Systems with Applications, 68, 151-162.

© IEOM Society International 771

Proceedings of the International Conference on Industrial Engineering and Operations Management
Pilsen, Czech Republic, July 23-26, 2019

Datta, S. (2012). Efficient Genetic Algorithm on Linear Programming Problem for Fittest Chromosomes. Journal of
Global Research in Computer Science, 3(6), 1-7.
Belea, R. (2004). The State Equation of A Genetic Algorithm. Journal of Control Engineering and Applied
Informatics, 6(3), 16-25.
Cuibus, O., & Letia, T. (2012). Cooperative Control achieved by Generic Genetic Fuzzy Logic. Journal of Control
Engineering and Applied Informatics, 14(3), 43-53.
Nicoara, E. S. (2007). Solving Scheduling In Pharmaceutical Industry with Genetic Algorithms. Journal of Control
Engineering and Applied Informatics, 9(1), 19-26.
Halal, F., & Dumitrache, I. (2006). Genetic Algorithm In Mobile Robot Control. Journal of Control Engineering
and Applied Informatics, 8(2), 21-30.

Biographies

Rahul sakharwade is a Junior Administrative Assistant, Director Secretariat, at All India Institute of Medical
Sciences, Raipur, Chhattisgarh, India. He earned B.E. in Mechanical Engineering from Chhattisgarh Swami
Vivekanand Technical University, Bhilai, Chhattisgarh, India, and Masters in Industrial Engineering and
Management from National Institute of Technology, Raipur India. He has published conference papers. His research
interest includes manufacturing, sequencing and lean manufacturing.

Udit Narayan Sahu is a Junior Research Fellow in the project financially assisted by Science and Engineering
Research Board under Department of Science and Technology, Govt. of India in the Department of Mechanical
Engineering at the National Institute of Technology Raipur, Chhattisgarh, India. He earned B.E. in Mechanical
Engineering from Chhattisgarh Swami Vivekanand Technical University, Bhilai, Chhattisgarh, India, and Masters in
Industrial Engineering and Management from National Institute of Technology, Raipur India. He has published
conference papers. His research interest includes additive manufacturing, micro electro-deposition, micro part
fabrication and 3D printing.

Mridul Singh Rajput is faculty in the department of Mechanical Engineering at NIT, Raipur, India and has
obtained his Ph.D degree from Indian Institute of Technology, Delhi, India. His research interest includes decision
algorithms and soft computing application under manufacturing domain belonging to conventional and non
conventional fields, Rapid Prototyping, ultrasonic assisted jet electrodeposition process and micro manufacturing.

© IEOM Society International 772